关键词: 神经网络, PID, 硬盘驱动器, 磁头定位

Abstract: In Hard Disk Drives（HDD），it is hard to establish accurate plant model.Aimed at this characteristic，this paper pre-
sents a method which applies neural network combined with PID controller.This scheme can realize real-time and on-line adjustment for proportional，integral and derivative parameters of PID controller using the capabilities of self-learning and arbitrary nonlinear expressions of neural network，as a result，contributes to finding the optimal combination of PID parameters so as to attain optimization of some performance target.As the modified style of conventional BP neural network，this paper uses the rule of minimizing total square error from initial time to current time as the performance function.The simulation results show that，applying this control scheme for HDD servo control，rapid convergent speed，short settling time，almost no overshoot as well as no steady-state error can be obtained.This method has superior to conventional one which is suitable for actual HDD magnetic head positioning.

Key words: Neural Network, PID, Hard Disk Drive, magnetic head positioning